Computational spoon shape optimization for use by patients with motor impairments

Sammanfattning: This thesis investigates whether the shape of eating utensils can be optimized using computational methods. This is done by creating a system to do so for a specific case, spoons, and see if any insurmountable problems are encountered. As this system is meant to optimize spoons rather than eating utensils in general, there are a few simplifications that can be done when modelling the eating process. The spoons are represented by a set of parameters that are further divided into groups. The first group representing the bowl, followed by a variable number of handle segments. The parameter sets are evaluated by calculating the volume between the lowest point on the rim and the bottom of the bowl, giving both the volume remaining at the end as well as the volume spilt over the course of the motion. This, together with the final direction and tilt of the bowl as well as a measure of spoon complexity, is processed into a score value. The chosen optimization method is gradient descent, with a variable step length. The results show that the optimization process does work, and the resulting spoons look plausible. The conclusion is that the approach is viable, but there are a number of changes and improvements needed to make things work for practical purposes.